What Is Confidence?

What if I told you that confidence might not make you better at something?

That we may have spent decades confusing cause and effect. That the person who performs well may not do so because they are confident, but may be confident because they have repeatedly performed well. And that, in some circumstances, increasing someone’s confidence without increasing their competence could actually make their judgement worse.

It feels counterintuitive because confidence occupies an almost unquestioned position in our culture. We tell children to believe in themselves. Athletes are encouraged to rediscover it when their form disappears. Leaders are expected to project it. Candidates are judged on it in interviews. If somebody struggles at work, one of the most common diagnoses is that they simply need to become more confident.

Entire industries have consequently developed around helping us acquire more of it. There are books, coaches, courses, affirmations, visualisation exercises and motivational speakers promising to help people overcome self-doubt and unlock the confident version of themselves apparently waiting underneath.

But there is a problem.

We rarely stop to ask what confidence actually is.

Is it a personality trait? A feeling? A belief about ourselves? A prediction about what will happen next? Is it something that causes good performance, or the psychological residue of having performed well before? Why can somebody be extraordinarily confident in one situation and deeply uncertain in another? And if confidence is inherently valuable, why are some of our worst decisions made when we are most certain that we are right?

Psychology offers a considerably more complicated picture than the one presented in most motivational literature.

Confidence appears to be less like a reservoir that some people possess in greater quantities than others, and more like an assessment being continuously constructed by the brain. It draws on previous experience, knowledge, feedback, emotional state and our beliefs about our own capabilities. Sometimes that assessment tracks reality extremely well. Sometimes it doesn’t.

Which raises a rather different question from the one we usually ask.

Perhaps the aim shouldn’t be to become more confident.

Perhaps the aim should be to become more accurately confident.

What do we actually mean by confidence?

One reason confidence is difficult to understand is that several different psychological ideas are routinely bundled together under the same word.

Self-esteem, for example, concerns our broader evaluation of ourselves. Optimism relates to expectations about future outcomes. Confidence can refer to certainty that a particular judgement is correct. Then there is self-efficacy, the concept developed most famously by psychologist Albert Bandura, which concerns our belief that we are capable of successfully performing a particular task.

That distinction matters.

A person can have relatively high self-esteem while having very low self-efficacy in a particular situation. You can broadly like yourself while being utterly unconvinced that you can deliver a speech to 500 people, negotiate a contract or parallel park with three impatient motorists waiting behind you.

Equally, someone can feel confident about one activity and deeply uncertain about another. Confidence therefore isn’t necessarily something that people simply possess. Much of it is contextual.

The American Psychological Association describes Bandura’s self-efficacy as a belief about our capacity to execute the behaviours required to produce particular outcomes. Bandura proposed four major sources for these beliefs: our previous performance, observing others, encouragement or persuasion from other people, and our interpretation of emotional and physiological states such as stress and anxiety.

More recent research into metacognition takes the idea further. Stephen Fleming, Professor of Cognitive Neuroscience at UCL, describes confidence judgements as inferential. The brain isn’t simply reading confidence from some internal gauge. It is constructing an assessment from information about the world, the decision in front of us and our own cognitive abilities.

This creates a much more interesting definition.

Confidence may be less a description of who you are and more a prediction your brain is making about what will happen next.

Perhaps confidence is a prediction

Imagine someone asks you a general knowledge question and you answer immediately.

You might say you are 95 per cent sure you’re right.

Another question arrives and this time you hesitate. You recognise the subject but cannot quite retrieve the answer. You guess and say you are only 55 per cent sure.

Nothing fundamental about your personality has changed between those two questions. What has changed is your brain’s assessment of the evidence available to it.

We make versions of these judgements constantly.

Can I do this? Do I understand this? Am I right? Will this work? Can I handle what happens if it doesn’t?

Most never arrive as neat numerical probabilities, but psychologically they perform a similar function. Confidence helps us decide whether to act, whether to seek more information, whether to reconsider an answer and whether to defer to somebody else.

Seen this way, a confident person isn’t necessarily someone who experiences less uncertainty. They may simply reach stronger judgements about how much uncertainty exists.

That distinction also explains why confidence can feel completely genuine while being completely wrong.

Your brain does not have direct access to objective reality. It has evidence about reality and a model of its own capabilities. If either is poor, the confidence judgement generated from them can also be poor. Fleming’s review makes precisely this point: because confidence depends partly on a person’s model of their own cognitive system, confidence and actual performance can diverge.

We experience the output, not the calculation.

A belief can therefore feel certain without being accurate.

Anyone who has confidently walked in the wrong direction after leaving a train station has conducted their own small experiment in metacognition.

Does confidence actually improve performance?

This is where things become particularly interesting, because much of the conventional wisdom around confidence assumes causation.

Someone performs brilliantly and we say they were confident. Someone performs badly and we say they lacked confidence. The obvious conclusion is that if we could somehow increase the second person’s confidence, their performance would improve.

There is certainly evidence linking confidence-related beliefs with performance.

A substantial 1998 meta-analysis by Alexander Stajkovic and Fred Luthans examined 114 studies involving more than 21,000 participants and found an average correlation of .38 between self-efficacy and work-related performance. People who believed more strongly in their ability to perform tasks also tended, on average, to perform better.

But correlation leaves us with an awkward question.

Which one came first?

If capable people repeatedly perform well, it would be entirely rational for them to become more confident. Their confidence may then contribute to subsequent behaviour, persistence or effort, but some of what we interpret as confidence producing success may simply be success producing confidence.

In 2013, Traci Sitzmann and Gillian Yeo looked specifically at this problem in a meta-analysis examining changes within the same individuals over time. Initially, self-efficacy and subsequent performance showed a positive relationship. Yet after controlling for the underlying trajectory of performance, the relationship between self-efficacy and future performance weakened substantially and was no longer statistically significant. Previous performance, by contrast, remained a significant predictor of subsequent self-efficacy. Their overall conclusion was striking: self-efficacy appeared to be primarily a product of past performance rather than the driving force behind future performance.

Sport provides another useful test because confidence is treated almost as an article of faith in elite performance. A 2022 meta-analysis examined 41 studies involving 3,711 athletes across 24 sports. The overall relationship between self-confidence and performance was positive, but relatively modest, at r = .25.

None of this means confidence is irrelevant. It would be equally simplistic to conclude that it has no effect at all. Believing that an action is possible can influence whether we attempt it, how much effort we expend, how persistent we are and how we respond when things go wrong.

But the causal story appears considerably more complicated than the motivational poster version.

High performers often have good reason to be confident.

Where does confidence come from?

Bandura’s framework gives us another important clue.

Of the four commonly discussed sources of self-efficacy, previous successful performance, usually described as mastery experience, tends to be particularly influential. A major review of the self-efficacy literature in education found mastery experience was typically the strongest source, although its importance varied depending on context and individual characteristics.

That makes intuitive sense.

Imagine that tomorrow you were required to give a talk to a thousand people.

If you had delivered hundreds of large conference talks before, your brain would possess considerable evidence about your ability to cope. You would know how you react when nervous, how to recover after losing your place, how to use the stage, how audiences behave and what happens when a joke doesn’t land.

Now imagine it was your first ever talk.

Somebody could spend the previous evening telling you how brilliant you are. You could repeat affirmations into a mirror. You could visualise a standing ovation.

Those things might alter your emotional state, and encouragement is one recognised source of self-efficacy. But none provides quite the same quality of evidence as having stood on a stage 200 times and survived.

This exposes something strange about the way confidence is often discussed. We treat it as a prerequisite for action when action is one of the mechanisms through which confidence is created.

People are frequently told to wait until they feel confident enough to do something. Psychology suggests they may sometimes have the sequence backwards. Experience provides evidence, evidence alters our expectations, and those expectations begin to feel like confidence.

This is also why good practice matters.

Not all repetition produces competence, and the popular claim that thousands of hours of practice inevitably create expertise has itself been overstated. A meta-analysis across multiple domains found deliberate practice explained meaningful but highly variable amounts of performance, from substantial proportions in games and music to less than one per cent in professional domains. Practice matters, but it is not everything.

The important distinction is that structured practice gives us something vague reassurance cannot: information.

We try something. It goes wrong. We receive feedback. We adjust. We try again. Eventually, situations that once contained enormous uncertainty become more predictable.

The subjective experience of that predictability is often what we call confidence.

The problem with too much confidence

If confidence helps us act, persist and influence others, the obvious temptation is to maximise it.

This is where the problems begin.

Consider what confidence is supposed to represent. If I say I am 90 per cent confident that a judgement is correct, ideally judgements made with that degree of confidence should turn out to be correct roughly 90 per cent of the time.

If they are correct 90 per cent of the time, my confidence is well calibrated.

If they are correct 55 per cent of the time, my confidence is not impressive. It is inaccurate.

The goal, therefore, cannot simply be maximum confidence. There are situations where uncertainty is the rational response.

Modern metacognition research distinguishes between how confident somebody feels and how accurately that confidence tracks their actual performance. Someone can possess enormous confidence while having relatively poor insight into when they are right and wrong. Conversely, someone can be highly skilled but systematically underestimate their own performance.

This is where the Dunning-Kruger effect inevitably appears, although probably not in the way it normally appears on LinkedIn.

Justin Kruger and David Dunning’s famous 1999 studies found that lower-performing participants in tasks involving humour, grammar and logic substantially overestimated their relative ability. The researchers argued that some of the skills required to perform well were also required to recognise good performance, creating what they described as a double burden for low performers.

The finding has since escaped academia and mutated into the much simpler idea that stupid people are too stupid to know they are stupid.

The research is more complicated.

Subsequent researchers have challenged how much of the familiar Dunning-Kruger pattern reflects a unique psychological deficit among poor performers and how much can emerge from statistical effects created by the way performance and self-assessment are measured and grouped. One 2020 analysis argued that much of the standard effect could be a statistical artefact and failed to reproduce the expected effect using alternative statistical methods on intelligence data.

The debate itself illustrates the broader point. Human beings are not particularly reliable instruments for measuring themselves, but nor can every instance of excessive confidence be explained by one neat psychological effect.

Sometimes we know less than we think we know. Sometimes we know more.

What matters is the size of the gap.

Why do we trust confident people?

There is another reason confidence matters, and it has little to do with whether the confident person is actually right.

Other people can see it.

Humans constantly make judgements about the competence of people whose competence we cannot directly observe. We select leaders, hire employees, follow experts and accept advice despite rarely possessing enough independent knowledge to verify everything somebody tells us.

Confidence becomes one of the signals we use.

Paul Price and Eric Stone demonstrated this in a series of experiments involving fictional financial advisers. Participants were presented with one reasonably well-calibrated adviser and another who expressed more extreme confidence. Despite the second adviser being overconfident, participants tended to prefer them. Greater confidence influenced perceptions of how knowledgeable the adviser appeared and how accurate people believed their judgements were. The researchers described this reliance on expressed certainty as a confidence heuristic.

Cameron Anderson and colleagues uncovered something equally fascinating when investigating confidence and social status. Across six studies, they found that overconfident individuals could attain higher social status because overconfidence produced behaviours that made them appear more competent to other people.

This creates an uncomfortable social asymmetry.

Competence can create confidence, but confidence can also create the appearance of competence.

Usually the two overlap enough for confidence to remain a useful signal. Someone speaking authoritatively about a subject may do so because they possess years of expertise. A surgeon who has performed the same procedure hundreds of times should probably sound more assured than somebody who read about it yesterday.

The difficulty arises when we confuse the signal with the underlying quality.

Certainty is easier to observe than expertise.

That may be why organisations can sometimes reward people who sound as though they know the answer over people sufficiently knowledgeable to recognise that the answer is uncertain.

Confidence is socially useful partly because humans need shortcuts.

It is also dangerous for exactly the same reason.

The illusion of knowing

One of the most revealing experiments in this area didn’t ask people to perform difficult professional tasks or predict financial markets. It asked them to explain ordinary objects.

In 2002, Leonid Rozenblit and Frank Keil investigated what they called the illusion of explanatory depth. People routinely believed they understood familiar mechanisms in far greater detail than they actually did. The illusion became apparent when participants were asked to produce detailed explanations of how those mechanisms worked. Their assessments of their own understanding subsequently fell.

It is a wonderfully simple observation.

Ask somebody whether they understand how a toilet works and the answer is likely to be yes.

Ask them to explain precisely how pressing the handle causes the tank to empty, refill and stop filling at the correct point and confidence can suddenly become less stable.

Knowledge often feels complete until reality asks us to use it.

The implication stretches far beyond household plumbing.

We can feel that we understand an economic argument until someone asks us to explain the causal mechanism. We can feel that we understand a strategy until somebody asks exactly how it will work. We can feel that we have mastered a skill until we attempt it under pressure.

Confidence can therefore survive quite happily in environments where it is never tested.

Explanation, practice and feedback are uncomfortable partly because they force the internal model to collide with external evidence.

That discomfort may be extraordinarily useful.

Maybe we shouldn’t be trying to build confidence

This leads to a slightly different way of thinking about confidence development.

Perhaps confidence itself shouldn’t always be the target.

If confidence is partly an inference based on previous experience, knowledge, feedback and our model of ourselves, then trying to modify the feeling directly risks working on the final link in a much longer chain.

We might be better off improving the evidence from which the judgement is made.

Build competence. Create realistic opportunities to practise. Give people useful feedback. Expose gaps between what they think they know and what they can actually do. Allow them to experience difficult situations repeatedly. Help them understand what good performance looks like. Make it safe enough to admit uncertainty so that uncertainty can be investigated rather than hidden.

Amy Edmondson’s research into psychological safety is relevant here. In her original field study of 51 work teams, psychological safety was associated with learning behaviours such as seeking feedback, discussing errors and experimenting. Those learning behaviours, in turn, were related to team performance.

This matters because cultures obsessed with appearing confident can inadvertently damage the mechanisms through which genuine confidence develops.

If admitting “I don’t know” is interpreted as weakness, people become better at concealing uncertainty rather than resolving it. If mistakes are embarrassing rather than informative, people defend them rather than examine them. If feedback threatens status, people avoid the evidence that might improve their internal assessment of their own capability.

The irony is that organisations trying to create confident people can end up creating people who are merely good at performing confidence.

There is a better ambition.

Not maximum confidence.

Calibrated confidence.

Knowing when you are likely to be right. Recognising when you may be wrong. Having enough experience to trust yourself when the evidence justifies it and enough self-awareness to become curious when it doesn’t.

That version of confidence is less glamorous than certainty.

It is also considerably more useful.

What does this mean for sellers?

Sales has an unusual relationship with confidence because it is both a performance requirement and a social signal.

A seller has to initiate conversations, ask difficult questions, challenge assumptions, navigate uncertainty, handle objections, discuss money, negotiate and occasionally tell a customer something they would rather not hear. A complete absence of confidence makes many of these behaviours difficult.

Research specifically within personal selling has therefore studied self-efficacy for decades, and reviews continue to find relationships between sales self-efficacy and sales performance, although researchers have also highlighted the importance of how self-efficacy is defined and measured. Context-specific assessments appear to tell us more than vague measures of whether somebody generally considers themselves confident.

That distinction is important for managers.

Saying that a seller “lacks confidence” tells us remarkably little.

Do they lack confidence opening a discovery conversation? Asking commercially sensitive questions? Speaking to senior executives? Challenging an unrealistic procurement demand? Presenting? Negotiating? Prospecting? Handling a particular objection?

Once confidence becomes specific, it also becomes coachable.

And the research suggests that the answer probably isn’t simply to tell the person to believe in themselves.

If mastery experience is one of the strongest contributors to self-efficacy, then confidence is likely to develop through successful experiences, including experiences deliberately created away from the customer.

This is one reason role-play is often misunderstood. Its purpose shouldn’t be to force sellers through awkward theatre for the sake of completing a training exercise. Done properly, practice allows people to encounter difficult situations before those situations carry commercial consequences. A seller can attempt the conversation, fail, receive feedback, adjust and try again. Over time, their brain acquires a much richer library of evidence about what happens when that situation occurs.

The goal isn’t merely to make them feel more confident.

It is to give them a reason to be.

Coaching performs a similar function when it is done well. The manager isn’t simply providing encouragement. They are helping the seller compare their internal assessment against external evidence.

This becomes particularly important at the other end of the confidence spectrum.

One of the most dangerous sellers in an organisation may not be the visibly nervous one. It may be the highly articulate, highly experienced seller whose confidence has stopped responding to evidence.

They are certain the opportunity is qualified because the customer likes them. They are confident there is budget because somebody mentioned a number three months ago. They believe they have a Champion because one contact takes their calls. They forecast the opportunity because they can construct a persuasive narrative explaining why it will close.

Confidence makes that narrative convincing to the seller and potentially to everybody around them.

Calibration asks a different set of questions.

What evidence do we actually have? What do we know? What are we assuming? How would we know if our assumption were wrong?

This is why good opportunity coaching should occasionally reduce confidence.

That might sound strange in an industry that has spent decades encouraging sellers to become more confident, but discovering that an opportunity is less healthy than you believed is not a coaching failure. It is an improvement in calibration.

The seller now possesses a more accurate model of reality.

Equally, coaching can reveal the opposite. Some capable sellers habitually underestimate themselves. Evidence from previous conversations, customer feedback and observed practice may show that their ability is considerably higher than the internal judgement they have constructed.

The manager’s job in both cases is not to increase confidence indiscriminately.

It is to close the gap between confidence and competence.

Perhaps that is the distinction we have missed.

We often talk about confidence as something people need before they can perform. Yet performance, practice, feedback and evidence are among the things from which confidence is built. Trying to manufacture the feeling without strengthening what sits underneath it can produce little more than bravado.

The best sellers are unlikely to be those who walk into every conversation convinced that they know exactly what will happen.

They are more likely to know what they are capable of, recognise what they don’t yet know, and remain sufficiently secure in that distinction to find out.

Perhaps, then, we have spent too much time trying to create confident people.

What we actually need are competent people whose confidence accurately reflects what they know.